🖥️ Teacher's Workbook: The 2-Hour Brain, on a Mac mini

Deep-dive class · 90 minutes · 19 slides · pairs with the projector deck · follows Build a Second Brain

This is everything behind the slides. The deck is what the room sees; this workbook is what you know. This is the power-user class — the room has (mostly) built their four folders in Class 2, and tonight you give that brain a body: a small always-on computer, a daily janitor, a weekly librarian, and one rule that keeps it all safe. Each segment below matches one slide with the goal, an expanded version of what to say, what to do with your hands, and the ways the moment goes sideways. Skim it the night before, sanitize your demo Inbox, print the worksheet, and teach off the deck.

Open the projector deck → ← All classes

THE SHAPE OF ITWhat this class does

People walk in with four folders and a nagging pile called "sort later." They walk out understanding a machine that sorts it for them — and holding the plain-English rulebook that machine will run on, written in their own handwriting. By the end, every person in the room should be able to:

What this class deliberately does not do: promise outcomes or pretend the machine is finished. AI does not cure, diagnose, or treat ADHD, and no study shows any AI tool improves ADHD outcomes. The automations are a roadmap you are installing in your own vault right now — you teach what you run, and you tell the room which parts are live as of class day. That honesty is why this class works with a skeptical room.

BEFORE YOU TEACHMaterials & setup

Bring:

The pre-class step you cannot skip — sanitize your demo Inbox. The 0:30 demo projects your real Inbox to a room of strangers. The night before: go through every capture you'll show, swap in fake names (Person A, Company B, Project X), and delete or move anything private — medical, legal, money, other people's business. Then seed it with five or six real-but-safe captures from your week: a voice-note idea, a saved article, a half-thought from 2am, something that obviously belongs in Projects, and one deliberately ambiguous one you can reject the filing proposal on. You're modeling the postcard rule with your own vault; if you wouldn't put a capture on a postcard, it doesn't go on a projector.

Room: tables beat rows — people write the rulebook by hand. This crowd skews returners from Class 2, but plan for a third of the room having no vault yet: the sprint is designed so they can play anyway. Recruit one helper for the sprint if you can get one.

Know your controls: arrow keys or clicking advances. Bullet slides reveal one line per press — you control the pace. The corner badge shows the run-sheet time so you always know if you're ahead or behind. The sprint timer has a 3-2-1 count-in and keeps counting even if you bump a key.

AT A GLANCEThe run sheet

ClockSegmentWhat's happening
0:00WelcomeFrame it: Class 2's brain gets a body and a pulse tonight
0:04Where we left offFour folders recap — and the "sort later" problem named
0:08The CODE loopCapture → Organize → Distill → Express, one reveal at a time
0:12The four layersVault · command center · helpers · capture — the whole machine
0:16Why a Mac miniAlways-on + files in your house — and "your laptop works too"
0:21The rules before we buildClass 2 first · roadmap honesty · never a silent move
0:25Janitor & librarianDaily triage and weekly sweep, introduced as characters
0:30Watch a triage happenLIVE DEMO — your real (sanitized) Inbox, one rejection on purpose
0:34The voice pipelinePhone → synced folder → the mini → filed by morning
0:38Privacy at the architecture levelYour hardware, postcard rule, fake names
0:42Pop quizRoom answers before the reveal: never a silent move
0:46Sprint (10 min)Everyone writes their janitor's rulebook — on paper
0:58Read me your rulesShare-out: one capture, where it goes — every rulebook differs
1:02The librarian's magic trickThree scattered notes become one drafted page
1:07The flywheelExpress feeds capture — the system compounds
1:12What it costsOne-time box vs. renting a brain monthly — honest numbers
1:16The honest limitsWeekend project, roadmap status, mistakes, no treatment claims
1:20Your first seven daysDay-by-day install plan, on the worksheet
1:26A brain with a bodyCODE one more time + confetti

SLIDE BY SLIDEThe segment guide

0:00 The 2-Hour Brain, on a Mac mini

Goal: set the premise — Class 2's brain gets a body tonight — and disarm the tech fear in the first minute.

Say

"Welcome back — and welcome, if this is your first one. Class 2 built you a second brain: four folders and a talking habit. Tonight we give that brain a body and a pulse — a little computer in your house that sorts, files, and summarizes your brain while you sleep. I want to kill one fear before it hatches: no coding background required. None. I'm a bartender — twenty-nine years behind a bar, not one behind a keyboard. I'm not an engineer; I'm the translator. Everything you'll see tonight is instructions written in plain English, the same English you'd use to tell a new hire where the glasses go. If you can explain your filing system to a person, you can explain it to this machine. That's the whole trick, and by the end you'll have written yours."

Do

If you brought the actual Mac mini, hold it up now — the "wait, that little thing?" reaction is the best opener you'll get. Make eye contact with anyone who looks like they wandered into the wrong room; they didn't.

Watch for

Visible tech dread — arms crossing at the word "computer." Name it out loud: "If the word 'automation' makes your shoulders climb, tonight's for you specifically. The sprint later is on paper." Shoulders come back down.

0:04 Where we left off

Goal: recap the four folders in ninety seconds and name the honest flaw — "sort later" never comes — so the whole night has a villain.

Say

"Quick recap for the returners, quick install for the new folks. The second brain is four folders: Inbox, Projects, Areas, Resources. One rule: dump messy into the Inbox, sort later. And everyone who's actually run it for a few weeks has noticed the catch — 'later' is doing a LOT of work in that sentence. Hands up if your Inbox has stuff in it from more than two weeks ago. Keep them up. Right — that's not a personal failure, that's the design meeting reality. An ADHD brain doesn't sort later. Sorting is boring, it's low-dopamine, and there's no deadline, which is the exact recipe for never. So tonight we're not going to try harder at sorting. We're going to hire something that sorts, every morning, whether we feel like it or not."

Do

Do the hands-up honestly and raise your own hand — your Inbox backs up too, and the room needs to know that. New folks with no vault: tell them to write the four folder names down now; that's their whole prerequisite.

Watch for

People who skipped Class 2 looking lost. Don't re-teach it — "four folders, dump messy, that's genuinely the whole prerequisite, and the paper you write tonight works whenever you build them." They stay in the room.

0:08 The loop the whole thing runs on

Goal: install the CODE loop — Capture, Organize, Distill, Express — and land the one 2026 twist: AI can run the two middle steps for you.

Say

"Here's the loop the whole machine runs on — four steps, borrowed from the Building a Second Brain world, and they spell CODE. Capture: everything lands in the Inbox — voice notes, web clips, the idea at 2am. You already do this. Organize: captures get filed to the right folder, tagged, linked to related notes. Distill: once a week, the pile becomes summaries and themes — the sense-making. Express: you make things from it, and what you make flows back in. Now notice something about those four steps. Capture is fun — that's the dopamine one. Express is fun — that's the making one. The two in the middle? Organize and Distill are the boring ones, the ones ADHD brains skip forever. And here's the 2026 twist this whole class hangs on: AI can now run those two middle steps for you. The fun stays yours. The boring gets hired out."

Do

Four reveals, one per press — say the word, then explain, don't rush. On the last line, physically point at the middle two words on the screen: "these two. These are what tonight automates."

Watch for

Someone who's read the Building a Second Brain book wanting to compare frameworks. Honor it fast — "same CODE, our four folders are the simplified ADHD cut" — and move; a methodology debate eats five minutes you don't have.

0:12 The four layers of the machine

Goal: give the room the full architecture in one picture — and make each layer feel like something they already half-own. This is the slide where overwhelm strikes; teach it bottom-up and slow.

Say

"Four layers, bottom to top, and I want you to notice how many you already have. The base: your vault — plain text files in those same four folders. No app owns them. Yours forever, readable on any computer made in the last forty years. Class 2 graduates: you own this layer already. The command center: an AI that can read and write those files. I use Claude Code — it talks like a chat app but works like an assistant who can actually open the filing cabinet. The helpers: small agent jobs that run on a schedule — you'll meet the janitor and the librarian in a few minutes. The capture layer: every door into the Inbox — phone dictation, notes, saved articles. You own most of this one too; it's your phone. So of four layers, most of you walked in with two. Tonight is about the middle."

Do

Reveal one layer per press, and use your hands — flat palm low for the vault, stack each layer on top. Bottom-up matters: starting from "plain text files, yours forever" grounds the scary layers in the boring one.

Watch for

This is the tech-overwhelm slide — you'll see eyes glaze at "agent jobs." Catch it live: "If that layer sounded like static, ignore it. You need exactly one sentence: small helpers that run on a schedule. The worksheet has the picture." Do not add detail to rescue a glazed look; detail is what caused it.

0:16 Why a Mac mini, specifically

Goal: justify the box honestly — always-on and files-at-home — while making equally clear that nobody needs to buy anything to start.

Say

"Why this box? Three honest reasons. It's the cheapest always-on computer Apple makes — a small box that sips power and sits under the TV. And always on is the point: the janitor works at 9am whether you woke up ready or not. That's the whole ADHD case — the system can't depend on you booting a laptop on a bad morning, because bad mornings are exactly when you need it. Second: your files stay in your house. That's the postcard rule built into the architecture — more on that in a few slides. Third, the honest cost: the box is a few hundred dollars one-time, plus your AI subscription. And now the part I need everyone to hear: you can run everything tonight teaches on the laptop you already own. The mini is for always-on. It is not the price of admission. Nobody should leave here thinking they have to spend money to start."

Do

If you brought the box, pass it to the front row — the weight of it makes "sits under the TV" real. Say the "laptop you already own" line twice; it's the line people mishear.

Watch for

"How much does it REALLY cost?" — asked with a squint. Answer with the full stack, nothing hidden: "The box, a few hundred one-time and optional. The AI subscription, monthly — real money, and the cost slide near the end puts it next to the alternatives. Sync tools, free. And the cost nobody lists: a weekend of your time to set up. That's the whole bill." The squint relaxes when the time cost gets named too.

0:21 The rules before we build

Goal: lay down the three honesty rules — prerequisites, roadmap status, and never-a-silent-move — before any demo makes the machine look magical.

Say

"Rules before we build, because this is the honest AI class and the rules come before the cool part. One: do Class 2's four folders first. Tonight is the upgrade, not the entry — a janitor with no house to clean is just a guy in your kitchen. Two, roadmap honesty, said plainly: I'm installing these automations in my own vault right now. This is not a finished product I'm selling you — it's a machine I'm building in my own house, and I'll tell you tonight exactly which parts are already live in mine and which are still on the bench. I teach what I run. Three, the rule that keeps all of it safe: never a silent move. The AI shows its filing plan and waits for your yes before anything moves. Your brain stays yours. And the standing rules: AI supports a brain — it does not treat, cure, or diagnose ADHD. And AI can be wrong, which is exactly why rule three exists."

Do

Four reveals, delivered flat and serious like the honesty slide in every class — contrast is credibility. Then actually tell them your live-versus-bench status as of today. Update that sentence every time you teach; a stale roadmap claim is a broken promise.

Watch for

Disappointment at "roadmap, not product" — someone came for a finished thing. Meet it head-on: "Would you rather I demo software I don't use? What you'll see at the demo is real and running. What isn't yet, I'll label. That's the deal here." Disappointment converts to trust when the demo is real.

0:25 Meet the janitor and the librarian

Goal: turn "scheduled agent jobs" into two characters the room can love — daily janitor, weekly librarian — and paint the Monday-morning goal state.

Say

"Meet the staff. The janitor works every morning around 9am — the Daily Inbox Triage. It goes through everything you captured, routes each thing to the right folder, tags it, links it to related notes — and shows you the plan before anything moves. Never a silent move; the janitor knocks. The librarian works Sunday night — the Weekly Sweep. It walks the shelves, finds orphan notes nothing links to, and spots themes: three notes on one idea becomes a drafted page. Then it writes you a week-in-review — here's what you captured, here's what kept coming up, here's what you might be circling without noticing. And the goal state, the picture I want in your head: you wake up Monday and your brain has already been tidied and summarized. Not by discipline. By staff. That's what we're hiring tonight."

Do

Give each character a small gesture — sweeping for the janitor, shelving for the librarian. Corny works; the names are doing memory work, same as YAP-MAP-TAP-SNAP did. "Tidied and summarized while you slept" gets a beat of silence after it.

Watch for

The first "why not just let it file automatically without asking?" — it often lands here, before the quiz. Don't spend the answer yet: "Hold that exact thought — it's the pop quiz in fifteen minutes, and I want you to get it wrong honestly." Say it warmly; you've just guaranteed a great quiz moment.

0:30 Watch a triage happen

Goal: the centerpiece. Project your real (sanitized) Inbox, run a live triage, and reject one proposal on camera — proving both that the machine works and that the human is the editor.

Say

"Enough theory — watch one happen. This is my actual vault, my actual Inbox, this week's actual mess. Fake names where names appear, because the postcard rule applies to me most of all. Six captures in the pile: a voice note, a saved article, a half-idea from 2am — you know the genre. I ask for a triage, in plain English, like this." [Run it. Read the proposals aloud as they appear.] "Look at the shape of this: it's a plan, not an action. Nothing has moved. It's proposing — this to Projects, this to Resources, tagged like so. Now watch the important part." [Reject one.] "No — that one's not filing where it thinks. I'm the editor. It adjusts, I approve the rest, and NOW it moves. That rejection you just watched? That's not the machine failing. That's the approval step earning its keep — and it's the reason I trust this thing with my brain."

Do

The night-before sanitize step made this moment safe — trust your prep and relax. Pick your rejection target in advance (the deliberately ambiguous capture you seeded) but let the reasoning be live. Narrate every beat; projected text is small and your voice is the real display. If wifi dies: walk through screenshots of a previous triage on the hotspot, and say that's what you're doing — honesty survives outages.

Watch for

The AI filing something genuinely weird, off-script. Best possible outcome — point at it: "See? Wrong, confidently. And nothing moved, because it has to ask. This is the whole safety model, demonstrated for free." Also watch for someone in front reading your projected captures too closely; if a capture draws the wrong kind of interest, keep moving — you sanitized for exactly this.

0:34 Capture from anywhere: the voice pipeline

Goal: connect the phone in their pocket to the box on the shelf — capture must be instant or it doesn't happen.

Say

"The machine at home is only half the system — the other half is the front door, and the front door is your phone. Here's the pipeline: you have a thought in the grocery store. You talk into a voice memo or a dictated note — five seconds, thumb and a sentence. That note drops into a folder that syncs. The folder syncs home to the mini. And in the morning, the janitor finds it and files it — you never touched a keyboard. Free sync tools do this between your phone and the box; that layer costs nothing. The capture law from Class 2 hasn't changed and never will: if it's not instant, it won't get done. An ADHD thought has a shelf life of about ninety seconds. Any capture that starts with 'open the app, find the button' loses the race. Talking is instant. Talking wins."

Do

Demo the front half live with your own phone: dictate a real capture at the mic — "note to self: the thing Person A said about the Tuesday class" — and tell them where it's now headed. The back half already showed itself in the triage demo; connect the two out loud.

Watch for

Someone stuck on which sync tool, wanting a brand-by-brand comparison. Keep it category-level: "Several free ones do it; the worksheet says 'set up sync' on Day 5, and any of them satisfies it. The pipeline shape is the lesson — tool names change yearly." Naming one winner from the front of the room ages badly.

0:38 Privacy at the architecture level

Goal: upgrade the postcard rule from behavior to architecture — and answer the "AI reading my files" fear before the quiz and sprint.

Say

"Let's talk about the fear in the room, because it's a good fear: 'this thing reads my files.' Here's what the architecture actually says. Your vault lives on your hardware — a box on your shelf, or the laptop on your table. That's not a feature we bolted on, it's the foundation the whole machine stands on. And the command center works under the rule you already know: never a silent move — it proposes, you approve, and you saw me reject one ten minutes ago. Now the part that still needs YOU: the AI chat window is not your vault. Any time you paste something into a chat, the postcard rule applies, same as Class 1. Fake names — Person A, Company B, Project X. No medical, no legal, no money details. Some things stay between you and humans, and the machine on your shelf never needs to know them to do its job."

Do

Four reveals, steady pace. Point to the worksheet's postcard box — it's printed there. If your setup sends vault content to a cloud AI service when it works, say so plainly here; the trust in this room is worth more than a tidy slide.

Watch for

The deep-privacy questioner who wants guarantees about AI companies. Same honest line as always: "I can't vouch for any company's policy and neither can anyone giving a talk. What I can give you is architecture — files on your hardware — plus the approval gate, plus the postcard rule for anything you paste. Three layers beats a promise." Don't get pulled into vendor terms-of-service exegesis.

0:42 Pop quiz

Goal: retrieval practice on the one rule that must survive tonight — the filing automation always shows its plan and waits.

Say

"Pop quiz — shout it out. What should the filing automation ALWAYS do? Option one: file things automatically without asking — that's the whole point, right? Option two: show its plan and wait for your approval before moving anything. Option three: delete old notes to keep the vault tidy." [Let the room commit before revealing.] "Two. Never a silent move. And I want to sell you on WHY, because option one sounds so efficient. The moment an automation moves files without showing you, you stop trusting your own brain — you open a folder and don't know if you put that there. And a second brain you don't trust is just a second pile, with extra steps. The approval tap costs you ten seconds a morning. The trust it buys is the entire product. Anyone who was holding the 'why not fully automatic' question from earlier — that was the answer, and you earned it."

Do

The quiz slide lets the room commit before the reveal — use the pause. If someone argues for option one, platform them for a moment; the room learning why "efficient" loses to "trusted" is the best two minutes of the night.

Watch for

The persistent automator: "I'd still want it silent — I don't want to approve things every day." Respect it, hold the line: "Run it supervised for a month first. If your rulebook gets so good the janitor is never wrong, you'll know exactly what you're choosing to trust. But you earn silent — you don't start there. Tonight, in this room, the rule is the rule."

0:46 Sprint: write your janitor's rulebook ⏱ 10:00

Goal: everyone leaves with the actual artifact — five real captures and where each goes, on paper. This paper IS the instructions they'll hand their AI.

Say

"Your turn, and it's paper on purpose — no logins, no wifi, no wrong app. Ten minutes on the clock. At the top of your page: your four folders. Then five REAL captures from your week — the voice memo you'd have taken, the article you saved, the 2am thought — and for each one, an arrow to where it should have gone. That's it. 'Podcast idea → Projects slash the channel.' 'Recipe from Marie → Resources.' Here's what I need you to understand about what you're writing: that list IS the instructions you'll hand your AI on Day 3. Word for word. You're not learning to code tonight — you're writing house rules in plain English, the same way you'd train a new hire. And if you don't have a vault yet? Write it anyway. The rulebook doesn't care when the house gets built. It ports."

Do

Start the timer — it has the 3-2-1 count-in — then WALK the whole ten minutes. Write your own rulebook visibly when you're not helping; body-doubling rules apply to teachers. The commonest fix: vague captures ("stuff about work") — push for the real one from their actual week. The worksheet has the five arrow rows printed; point people there.

Watch for

No-vault people frozen at the folder line — sit with them for thirty seconds: "Four words: Inbox, Projects, Areas, Resources. Write them, done, now the fun part." Perfectionists designing elaborate taxonomies with sub-sub-folders: "Five captures, four folders. If your rulebook needs a rulebook, the janitor quits." Fast finishers: "Add capture six — the weirdest thing you'd ever save. Ambiguous ones train the best rules."

0:58 Read me your rules

Goal: share-out that proves the point — every rulebook is different, and different is correct. The machine adapts to the brain, not the reverse.

Say

"A few volunteers — read me ONE line from your rulebook. One capture, and where it goes. Sharing's optional, always." [Take three or four. After each, reflect it back.] "Did you catch what just happened across those? Same exercise, and no two rulebooks match. One of you files recipes under Resources, one under a cooking project, and one of you — beautifully — has a folder that's basically 'things Future Me will want.' Nobody's rulebook is wrong. That's not a nice sentiment, that's the design spec: the machine adapts to YOUR brain, not the other way around. Every filing app you've ever abandoned made you learn its categories. This is the first system where the categories learn you. That page in your hand is the most personal piece of software instruction you'll ever write, and you wrote it in ten minutes with a pen."

Do

Count five silently after asking — someone always breaks the silence. Thank every share by name. If nobody volunteers, read a line from YOUR rulebook and ask "who files that somewhere different?" — disagreement is participation, and it makes the point even faster.

Watch for

Someone hearing another person's rulebook and deciding theirs is wrong — you'll see the eraser come out. Stop it publicly and kindly: "Don't you dare copy hers. Her janitor works for her brain. Yours works for yours. That's the whole slide."

1:02 The librarian's magic trick

Goal: sell the weekly sweep with its best moment — three scattered notes coming back as one drafted page. The vault starts writing back.

Say

"The janitor keeps the house clean. The librarian is where it gets spooky — in the good way. The weekly sweep reads across everything, and here's its signature move: three scattered notes about the same idea — one from a Tuesday yap, one from an article you clipped, one from a 2am thought you don't even remember having — come back Sunday night as ONE drafted page, with the connections already made. You didn't know those three notes were about the same thing. It did. And for a brain like ours, that matters more than tidiness ever will, because ADHD brains are pattern-making machines with no pattern-storage. We SEE connections constantly — in the shower, on the drive — and lose them by lunch. The librarian is pattern-storage. Your vault stops being a filing cabinet and starts writing back to you. That's the moment this stops being organization and starts being a second brain."

Do

If your own librarian has produced a real week-in-review by class day, read two lines of it (sanitized) — nothing sells this slide like a real artifact. If it hasn't yet, say so — roadmap honesty — and describe the design instead. Never fake the artifact.

Watch for

"Is this the part that's live in your vault, or the roadmap part?" — the room remembering your 0:21 promise is a GOOD sign. Answer with today's true status, whatever it is. This question is the trust you built coming back to check on you; pay it in full.

1:07 The flywheel: express feeds capture

Goal: close the CODE loop — what you make goes back in, so the system compounds instead of just accumulating.

Say

"Last piece of the loop, the one that turns a filing system into a flywheel. Everything you MAKE — a post, a script, an email you're actually proud of — goes back into the vault. That's Express feeding Capture. Why it matters: next month's brain has this month's wins in it. When you sit down to write anything, the librarian can hand you what worked last time, in your own voice. The system compounds — most systems just accumulate. And one data point for scale, honestly labeled: Meta built an internal AI second brain on this same folder philosophy, and by their own account it reached 63,000 installs in three months. I can't verify their number — 'by their own account' is the label. But here's the part I can verify: yours doesn't need 63,000 users. It needs exactly one, and that user is the reason we're all here."

Do

Draw the loop closing with your finger — Express curling back into Capture. Keep the Meta stat exactly as worded: "by their own account," nothing firmer. The pivot from 63,000 to "one very important user" is the landing; give it a beat.

Watch for

Non-creators opting out: "I don't make content, so this isn't me." Widen it: "An email you got right, notes from a hard phone call, the packing list that finally worked — that's all Express. If you'd ever want to find it again, it feeds the wheel."

1:12 What this costs vs. renting a brain

Goal: the money slide, played straight — one-time box versus monthly rented brains, prices labeled with their source and their shelf life.

Say

"The money slide, and we play it straight. This setup: a one-time small computer, plus an AI subscription, plus free sync tools. The subscription is the real recurring cost — I won't hide it. Now the rented brains. Notion AI: twenty-plus dollars a month, forever, with the full AI behind the Business tier. Mem: twelve to ninety-nine a month. Tana: twenty to a hundred twenty a month. Source and shelf life: vendor pages, July 2026 — prices drift, so check before you quote me. But the deeper difference isn't the arithmetic, and it's the reason I'd teach this class even if the monthly tools were free: their files live in their cloud. Yours live on your shelf. If any of those companies pivots, sells, or sunsets — and tools do — your plain text files on your own hardware don't even notice. You can rent a brain. I'd rather you own one."

Do

Read the prices exactly as the slide has them, with the "vendor pages, July 2026 — prices drift" label said out loud, not skipped. Do not add tools or prices from memory — the slide is the whole approved list.

Watch for

"So which subscription do you pay, and how much?" — fair question, answer it honestly for your own setup and label it as yours: "That's my bill, today, for my usage — yours will differ and prices drift." Don't turn your personal number into the class's official price; the slide's sourcing rule exists so nobody quotes you a stale figure in six months.

1:16 The honest limits

Goal: the trust bookend — say every limitation plainly before the send-off, so nobody leaves oversold.

Say

"Before the pep talk, the fine print — read aloud, because that's the house style. One: setup is a weekend project, not two hours. Budget for that honestly, or the half-built machine becomes another abandoned project on the pile, and we have enough of those. Two, one more time so nobody misses it: the automations here are a roadmap I'm installing in my own vault right now. I teach what I run, and you heard tonight which parts are live in mine as of today. Three: agents make mistakes. They file things weird — you watched mine try. The approval gate isn't training wheels you'll outgrow; it's the design. Four, the one that never gets skipped in any class I teach: none of this treats ADHD. Not the box, not the janitor, not the librarian. What it does is catch what your working memory drops. That's the whole promise — and honestly? For brains like ours, that's a lot."

Do

Slow, flat, serious — the same register as every honesty slide. If you're running behind, this slide does NOT get cut; trim the flywheel or the librarian instead. The warnings are the brand.

Watch for

Relief, again — this room has been to hype talks before, and "weekend project, agents make mistakes, doesn't treat anything" is not what they braced for. That exhale is repeat attendance being born.

1:20 Your first seven days

Goal: convert tonight into a day-by-day install plan — one small step per day, ending with a triage they run themselves.

Say

"Homework, ADHD-sized: one step a day, and it's all on your worksheet with checkboxes. Day one — is your Class 2 vault built? Good; pick the machine. The laptop you own, or order the mini. Owning nothing new is a valid Day 1. Day two, install the AI command center and let it read one folder. One. Day three — take out tonight's paper and hand it your janitor rulebook. You already did the hard part with a pen. Day four, first supervised triage: approve or reject every single move, like you watched me do. Day five, phone-to-computer sync, so the grocery-store thoughts make it home. Day six, ask for your first week-in-review. Day seven, run the triage yourself with one message. And when you eventually graduate it to a schedule, that step starts happening while you sleep. The morning you wake up and it's already done — that feeling is the product."

Do

One day per press, but briskly — the detail lives on the worksheet, so this is a flyover. Land hard on Day 3: point at the actual papers in their actual hands. "That page. That's Day 3, already done."

Watch for

The overwhelmed no-vault attendee looking at a seven-day plan they can't start. Reroute them warmly: "Your Day 1 is the four folders — thirty minutes, and Class 2's materials are free on the site. The rulebook in your hand is patient." Overachievers planning the whole week for tomorrow: "One day at a time. The mini isn't going anywhere; it literally sits still for a living."

1:26 A brain with a body

Goal: end on the CODE loop said together, the rulebook in their hands as proof, and confetti.

Say

"Look at what you're holding: your own janitor's rulebook, in your own handwriting, for your own brain. You didn't watch a class about a machine tonight — you wrote the instructions it runs on. Four words, one more time, all together: Capture. Organize. Distill. Express. You keep the fun ones — the capturing, the making. The machine takes the boring middle, and it asks permission before it touches anything, forever. Here's the send-off: one day at a time, a little each day — and now the little box does its share while you sleep. Some weeks all you'll manage is capture. A capture-only week is still a week nothing got lost, and for brains like ours that's already a different life. Thank you — I'll stick around for questions, and everything from tonight lives free at adhdyap.com."

Do

The four words land one per press with the room saying them; the slide fires confetti on the last. Then stop talking and let it land. Stay 15 minutes after — the setup questions arrive one-on-one, and so do the people who were too shy to say they never built the vault.

Watch for

The host — thank them by name in front of the room. And catch anyone leaving without their rulebook page on the table; chase them down with it. That paper is the class.

BE READYTough questions from the room

"I don't want an AI reading my files. How is this okay?"

"Three layers, and you saw all of them tonight. Layer one: the vault lives on your hardware — a box on your shelf or the laptop you own — not in someone's cloud. Layer two: never a silent move — the AI proposes and you approve, and you watched me reject a proposal live. Layer three: the postcard rule for anything you paste into a chat window — fake names, no medical, legal, or money details. Three layers beats any company's promise. And the honest add: nothing sensitive ever NEEDS to go in for the system to work. The janitor files 'call Person A back' just fine without knowing who Person A is."

"What does this REALLY cost, all-in?"

"Full stack, nothing hidden: the box is a few hundred dollars one-time and optional — your current laptop runs everything. The AI subscription is the real recurring cost; check current prices because they drift. Sync tools: free. And the cost nobody advertises: a weekend of setup time. Compare that to the rented brains on the cost slide — Notion AI at $20+ a month, Mem at $12–99, Tana at $20–120, per their vendor pages as of July 2026 — and the one-time-versus-forever math tells its own story. But do the weekend honestly or don't start; a half-built machine is just another abandoned project."

"I skipped Class 2 and don't have a vault. Was tonight wasted on me?"

"Not even slightly — you wrote the rulebook, and the rulebook is the personal part. The vault is four folders you can build in thirty minutes: Inbox, Projects, Areas, Resources. Class 2's materials are free on the site, no email wall. Build the folders this week, and the paper from tonight ports straight in — it doesn't care when the house gets built. You did tonight's hard part with a pen."

"Why can't it just file everything automatically? The approvals sound annoying."

"That was the quiz, and the answer is the hill this class dies on: never a silent move. The moment files move without you seeing the plan, you stop trusting your own vault — you open a folder and don't know if you put that there — and a second brain you don't trust is just a second pile. The approval costs ten seconds a morning. If after a month of supervised triage your janitor is never wrong, you'll know exactly what you'd be choosing to trust. You earn automatic; you don't start there."

"Do I have to buy a Mac mini? I can't spend that right now."

"No, and please don't buy one tonight. Everything in this class runs on the laptop you already own — the mini buys you exactly one thing: always-on, so the janitor works whether you booted up or not. Run it on your laptop for a month first. If you find yourself wishing the triage had happened before you woke up, that wish is the purchase decision, and it'll be an informed one."

"Is this actually running in your house right now, or is this theoretical?"

"Fair question, and it deserves today's answer, not a script: I'm installing these automations in my own vault right now, and at the rules slide I told you which parts are live as of class day. The triage you watched was real — my vault, my mess, my rejection. What isn't live yet, I label as roadmap. I teach what I run, and the roadmap shrinks every time I teach this class. If it ever stops shrinking, I'll say that too."

"Will this actually help my ADHD? Where's the evidence?"

"Honest answer, same as every class: AI does not cure, diagnose, or treat ADHD, and no study shows any AI tool improves ADHD outcomes — including this setup. What this system does is narrower and real: it catches what working memory drops. Captures stop vanishing, patterns stop dissolving, Monday starts sorted. That's support, not treatment. Run it for a month and count what stopped getting lost — your own data is the only evidence that matters for your brain."

"What happens to my files if the AI company disappears or I stop paying?"

"This is the best question in the room, because the answer is the architecture: your vault is plain text files, in folders, on your hardware. If the AI subscription ends, the janitor stops showing up — but the house and everything in it stays yours, readable by any computer, forever. Compare that to a cloud tool sunsetting with your notes inside it. You lose the staff, never the brain. That's the whole reason the base layer is boring on purpose."

PRINT MEStudent worksheet — The 2-Hour Brain, on a Mac mini

Print one per seat (this section prints clean on its own — use the print button up top). It's the take-home version of the whole night, including the rulebook they write in the sprint.

🖥️ The 2-Hour Brain, on a Mac mini — my worksheet

Name: ____________________ · Date: ____________

My four folders (from Class 2 — write them here)

1. ______________________ (everything lands here first — dump messy)

2. ______________________ (active things with a finish line)

3. ______________________ (ongoing parts of life, no finish line)

4. ______________________ (reference — stuff Future Me will want)

The CODE loop (one line each, as we go)

Capture = ______________________________________ (everything into the Inbox, instantly)

Organize = _____________________________________ (filed, tagged, linked — the janitor's job)

Distill = _______________________________________ (weekly summaries and themes — the librarian's job)

Express = ______________________________________ (make things; what I make flows back in)

🧹 My janitor's rulebook (the 10-minute sprint)

Five real captures from my week → where each one goes. This list IS the instructions I'll hand my AI on Day 3.

1. Capture: ______________________ → goes to: ______________________

2. Capture: ______________________ → goes to: ______________________

3. Capture: ______________________ → goes to: ______________________

4. Capture: ______________________ → goes to: ______________________

5. Capture: ______________________ → goes to: ______________________

🛡️ The automation rule

Never a silent move. The AI shows its filing plan and waits for my approval before anything moves. The approval step isn't training wheels — it's the design. A second brain I don't trust is just a second pile.

📮 The postcard rule (still the law)

The AI chat window is not my vault. Nothing gets pasted in that I wouldn't write on a postcard. Fake names only — Person A, Company B, Project X. No medical, legal, or money details. And: AI can be wrong. That's exactly why the approval step exists.

📅 My first seven days

  • Day 1 — Class 2 vault built? Then pick my machine: the laptop I own, or order the mini
  • Day 2 — Install the AI command center and let it read ONE folder
  • Day 3 — Hand it my janitor rulebook (the one on this page)
  • Day 4 — First supervised triage: approve or reject every move
  • Day 5 — Set up phone-to-computer sync for voice captures
  • Day 6 — Ask for my first week-in-review
  • Day 7 — Run the triage myself with one message

Everything from tonight, free: adhdyap.com · AI supports an ADHD brain. It does not cure, diagnose, or treat ADHD — and it can be wrong. Setup is a weekend project — budget honestly.

AFTERWARDAfter class